The specific aim of this study is to document observable changes in confused, elderly client's wandering and wandering-related behaviors as a result of participation in a physical exercise program. The quasi-experimental design will solicit 30 long-term care residents in a VA facility, who meet eligibility criteria, to participate in multiple pre- and post-tests to measure effectiveness of two variations (indoor and outdoor) of a single treatment, physical exercise. Following a 3-month baseline period, the total sample will be randomly assigned to the two treatment variations (N = 15 each), with assignments to be reversed after 12 weeks. Differences in mean scores for each subject and group, pre, during, and post treatment will be reported to provide the following comparisons: Indoor exercise vs. routine care; outdoor exercise vs. routine care; and indoor vs. outdoor exercise. Statistical significance will be tested using repeated measures analysis of variance.